47, NO. 6, DECEMBER 2000. 1781. NEW LSO BASED SCINTILLATORS. J.S. Iwanczyk', Senior ... due to its high density and high effective atomic number. It is.
IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 47, NO. 6 , DECEMBER 2000
1781
NEW LSO BASED SCINTILLATORS J.S. Iwanczyk', Senior Member, IEEE, B.E. Patt', Member, IEEE, C.R. Tull', Member, IEEE, L.R. MacDonald', Member, IEEE, Eric Bescher192,S.R. Robson2,J.D. Mackenzie2,and E.J. Hoffman3,Fellow IEEE 1
Photon Imaging, Inc, 19355 Business Center Drive, Suite 8, Northridge, CA 91324 2UCLADepartment of Materials Science 3Division of Nuclear Medicine & Biophysics, Department of Molecular & Medical Pharmacology, UCLA School of Medicine, Los Angeles, CA 90095
Abstract Cerium-doped lutetium oxyorthosilicate (LSO) is amongst the most promising new scintillators discovered in almost five decades, with a unique combination of important properties for x and gamma-ray spectroscopy, namely: high density, fast decay, and large light yield. LSO seems to be a prime candidate to replace BGO in PET systems. However, the practical utilization of LSO is hindered by difficulties related to crystal growth (Czochralski method) due to the high temperatures employed. A new approach has been developed using a low-temperature crystal growth technology to produce scintillating LSO crystals. Light transparent polycrystalline LSO samples of a few mm3 in volume were grown and characterized by XRD, optical absorption, light decay measurement and gamma-ray spectral response. The properties of the new crystals compared well with highquality crystals grown by the Czochralski method. I. INTRODUCTION In 1948, Robert Hofstadter [l] first demonstrated that crystalline sodium iodide with a very small amount of activator (thallium) NaI(T1) produced an exceptionally large scintillation light output in response to gamma radiation compared with the organic materials that had been previously tried. This discovery opened the era of modern scintillation spectrometry of gamma and x-ray radiation. Remarkably, this same material: NaI(T1) remains dominant in many application areas despite almost five decades of subsequent research devoted to other scintillating materials. It remains of paramount importance today because it has a large light output and it is relative easy to grow large volume ingots of NaI(T1). However, it also has many serious deficiencies, which in some cases limit the performance of instruments that use it, and in other cases simply prevent it from being used at all. Thus there has been a constant interest in the development of new inorganic scintillators that would have significantly improved properties, compared to NaI(Tl), for applications such as medical imaging, high (and low) energy physics, industrial process control, monitoring of environmental pollution, geophysics, and space exploration. Table I summarizes the properties of common as well as lesser-known recently introduced inorganic scintillators. A significant advance was achieved with the introduction of bismuth germanate (BGO). BGO has a very high detection efficiency 0018-9499/00$10.00
due to its high density and high effective atomic number. It is very rugged and non-hygroscopic, but has relatively low light output and even longer decay time of the scintillating light than NaI(T1). Early Positron Emission Tomography (PET) systems employed NaI(Tl), which was soon replaced by BGO due to its high photoelectric cross-section (4 times larger at 5 11 keV). The most promising scintillator, which was first introduced within the last several years, is cerium-doped lutetium oxyorthosilicate (LSO). These new crystals appear to have outstanding scintillating properties. The material shows an unique combination of high density, a fast scintillation decay time, and a large scintillation light yield relative to that of NaI(T1). In addition, it is not hygroscopic and it is reasonably rugged, resulting in large convenience for handling and packaging. The combination of the above properties, which are extremely attractive for a large variety of applications, have led to a high degree of excitement in the scientific community. One of the most interesting applications involves the replacement of BGO with LSO crystals in Positron Emission Tomography leading to the construction of novel more p o w e h l medical diagnostic systems. Unfortunately, the practical utilization of this material is hindered by difficulties of high temperature growth of reasonably large size crystals with an uniform high light output. TABLE I: Properties of Inorganic Scintillators (g/cm3)
NaI(T1)
3.67 4.51 4.5 1 4.08 4.89 4.89 3.19 7.13 6.7 5.55 4.55 8.34 7.4
CsI(T1) CsI(Na) LiI(Eu)
aBaFz bBaF2 CaF2(Eu) BGO
GSO YAP YAG LuAP LSO
I
1
(nm)
I
415 540 420 470 310 220 435 505 440 365 520 365 420
, 0.0006 1.44
1.82
I 30.040
HY
(PmeV)
38,000 52,000 39,000 11,000 10,000 1,400 24,000 8,200 7,620 18,000 16,700 11,300 27,300
Densitv?Wavelennth of Maximui Emission. Index of Refraction 4Princi& Decay C&stant>Total Light Yield (Photons/MeV) 6Hygroscopic: Yes/No, 'slow comp., bfast comp. 0 2000 IEEE
1
5y
2L"
IP
Material
N
N N
1782
Initially, LSO crystals were grown by Schlumberger-Doll Research employing the Czochralski technique [2-71. The Czochralski method was specially selected as it is generally thought to be an appealing method for growing crystals of high-melting-point (1 900 - 2 100°C) oxide materials such as LSO. Due to the use of very high temperature the crucible material has been restricted to iridium imposing severe growth limitations. Recently, the growth technology has been transferred from Schlumberger-Doll Research to CTI, Inc. with the promise of finalizing of the process [8]. Also, there are other attempts to grow LSO crystals all over the world [9,101. Our approach shown in this paper represents a complete departure from traditional methods of growing single crystal scintillators. The paper presents methods for producing transparent bulk polycrystalline scintillators composed of small size (smaller than the scintillation light wavelength to avoid scattering at the grain boundaries) optically transparent polycrystalline grains. A key innovation is the use of the solgel process, which involves the reaction of liquid organometallic precursors at room temperature. This technique has been used with success in the production of many glasses and ceramics [ll], polycrystals and single crystals [12]. Although many ceramics have been fabricated via the sol-gel technique, production of lutetium orthosilicate (LSO) via sol-gel has not yet been investigated. The presented method offers many advantages over conventional techniques, such as low processing temperatures (500 1200°C), controlled stoichiometry, homogeneity at the molecular level, which is particularly important if a high and homogenous dispersion of the dopant is desirable, high purity, possibilities of custom composition and shape of crystals, and low processing costs. THESOL-GEL PROCESS We have explored several routes for the preparation of LSO crystals by the gel method, which eventually led to the scintillating materials described in this paper. The synthesis routes used commercially available materials such as Lutetium nitrates or acetate, or Lutetium alkoxide synthesized in the laboratory. The best results have been obtained by the synthesis of the alkoxide directly from reaction of a Lutetium metal with alcohol in the presence of a catalyst. Lutetium isopropoxide was obtained by refluxing at 82°C of 40 mesh Lutetium metal powder (Alfa Chemical, 99.9%) with anhydrous isopropanol (Aldrich Chemical, 99.9%) under dry nitrogen. The reaction has been carried out for more than 24 hours. The catalyst used, HgC12, is believed to increase the reaction rate by forming an amalgam with the metal on its surface, which reacts more easily with the alcohol. The presence of -40 ppm of water would initiate hydrolytic decomposition of the alkoxide. Therefore, extreme care has been taken in the synthesis in order to avoid contamination with air or moisture. After 3-5 days a mixture of TEOS, 1% by weight cerium (111) nitrate and anhydrous isopropanol has been injected directly into the cooled lutetium alkoxide mixture under a strong flow of nitrogen at room temperature. The mixture
was refluxed for about 4 hours at room temperature. The resulting lutetium-silicon solution was centrifuged to yield a transparent solution. Gel synthesis from these solutions was investigated using different techniques. In one method, water at pH 1 was added after casting, thus speeding up the gelation time. In another method, the solution was left to slowly react with ambient moisture. The addition of water seemed to promote precipitation of a hydroxide in the gel, whereas samples, which had slowly adsorbed moisture from the atmosphere, yielded transparent gels. The yields of the initial reaction were low but improved for the more recent batches. Current yields of about 60% can still be improved in the future. A flow-chart of the process is shown in Figure 1. The samples were fired at 1200°C for 1 hour, before being oven cooled. The crystallization occurred at 1087°C as measured by differential thermal analysis. As a result transparent sol gel crystal samples with sizes ranging from 1.5 x 2 x 0.85 mm3 to 1.6 x 6.1 x 0.75 mm3 have been obtained. The transparent crystals were composed of small Lu silicate polycrystalline grains suspended in the transparent Si02 matrix. Several LdSi ratios were explored. The LdSi ratio in the samples was not optimum. The highest value of the ratio was 4. Therefore, future efforts will focus on increasing the LdSi ratio and increasing the density of the crystal. As indicated by diffraction measurements during experimentation with LdSi ratios we have been able to obtain two forms of transparent crystals of Lutetium silicates namely oxyorthosilicate (Lu$3iOs) (LSO) and pyrosilicate (Lu2Si2O.r). Lutetium pyrosilicate showed very similar scintillating characteristics to LSO and the generated results will be the subject of a separate paper.
11.
1
I
I
I
I Reflux 3-5 days I
Cast gel
1
I q i Fire I
Figure 1. Flow chart of sol-gel processing of LSO crystals
1783
CRYSTALS 111. CHARACTERIZATION OF THEGROWN The properties of sol-gel grown LSO crystals were evaluated by variety of methods including diffraction and light absorption. The scintillating properties have been tested in response to different gamma-ray sources and compared with those obtained from crystals grown by standard Czochralski method pulling from the melt.
A. Diffraction Measurements Samples were analyzed by TEM with an EDX attachment. EDX analysis confirmed the presence of Lu, Si and Ce dopant in the material. Observation of the microstructure of the crystal revealed the presence of amorphous silica, and small, angular shaped crystalline lutetium orthosilicate crystals with a monoclinic structure. The XRD on these crystals confirms the monoclinic structure (Figure 2), as does the electron diffractionpattern (Figure 3). ID aneleccyslal 'ile PHOTON MDI
Scan 10-80/04/1/#1751,Anode CU
optical coupling grease (Bicron BC-630). Measurements (I) of the spectral response of a Czochralski grown crystal (2.0 x 1.7 x 2.5 mm3) and two sol-gel grown samples (each 1.6 x 6.1 x 0.75 mm3) were made. A photograph of representative samples is shown in Figure 4. The Czochralski crystal appears to be more transparent in the photograph. This is somewhat due to its flat, parallel, and highly polished surfaces as contrasted to the as-grown surfaces of the sol-gel crystal. On the other hand, the light transmission of the sol-gel crystal appears to be better than that of the Czochralslu crystal as measured in the transmission spectra (VIo) of sol-gel and Czochralski LSO crystals as a function of wavelength shown in Figure 5. In fact we believe that the transparency of the crystals produced by the two methods are very similar and very high (>95%), and that the discrepancies in Figure 5 are anomalous, being due to the variant preparation of the surfaces of the samples and resulting differences in the amount of reflected light.
zero4 0
1
!I
P-Theto ;., dl .0?53.tu:'
$1
115.Lutetumsilicora
Figure 2- XRD of a LSO sample
sol-gel Figure 4: LSO crystal (left) and 1.7~2.5~1.5"~ Czochralski LSO crystal (right).
-E
100%
80%
-m-Czochralski
0%
I
*
I .
400
Figure 3- Electron diffraction of a LSO sample.
B. Optical Absorption Optical absorption was measured using a tunablewavelength monochromator system. All measurements were made using a NIST traceable diode (UDT model PIN lODP, SN80256). First the spectral response of the diode alone (IO) was measured. Then each of the test samples was coupled directly to the optical window of the photodiode using an
..
1 1 .
I
I..
LSO (1.7x1.6x2.5mm3)
I I . 1 I . I
I
*
I I I .I
..
I
I
I . I. I . . 1
420 440 460 Wavelength (nm)
480
Figure 5: The transmission (IDo) spectra of sol-gel and Czochralski LSO crystals as a function of incident wavelength. C. Light output measurement
The light output of the gel samples was measured by coupling one face of the crystal to the PMT using grease. Because the crystals are transparent, all other faces of the samples were covered with Teflon tape to reflect light and allow for the maximum of light to be directed toward the
1784
cathode of the PMT. A Cs-137 gamma source (662 keV) was used to generate light signals in LSO crystals. The corresponding electrical signals from the phototube were then fed to a Hamamatsu preamplifier and a Tennelec Tc-244 spectroscopy amplifier. Shaping time of the spectroscopy amplifier was between 1 - 3 ps in order to be several times longer than the decay time of the scintillator. The amplitudes of the pulses were then digitized and histogrammed with a NucleusTMmultichannel analyzer (MCA). The peak position corresponding to 662 keV was determined for the various samples and compared with the peak position observed for 662 keV in Czochralski grown crystals in order to allow assessment of the relative light output. An effort was made to select similar sizes of the studied crystals. The Czochralski grown crystal was pre-selected for high light output. A comparison of the energy spectra for Cs-137 (662 keV) collected from a Sol-Gel crystal (1.5 x 2 x 0.85 mm3) and a Czochralski crystal (2.0 x 1.7 x 2.5 mm3) is shown in Figure 6. The system gain was the same for both data acquisitions so that the difference in the peak position (number of channels) is directly proportional to the difference in light output between the two samples. The ratio of the light output of the sol-gel to the Czochralski was 65.5%.
0
50
100 150 200 250 300 350 400 Channel
functions and the decay times shown in Fig. 7 were extracted from the fit parameters. Three sol-gel LSO samples were measured. The decay times of these three sol-gel samples were 34.4, 34.9 and 34.7 ns. 0
10
20
time (ns) 30
40
50
60
70
80
90 100 I
-
-
decay time (ns;
Sol-Gel LSO 34.7 Czochralski LSO 41.1
Plastic
2.18
Figure 7. Digitized PMT pulses from LSO samples and a plastic scintillator.
The 2.18 ns result for the plastic scintillator confirms that the RCA PMT is fast enough so that system limitations do not affect the LSO measurements. Although the most commonly quoted decay time for Czochralski grown LSO is 40 ns, different batches are observed to have a range of decay constants. A recent report on a large database of Czochralski LSO gave a decay time range of 28 to 46 ns with an average of 37 ns [13]. This variation is not entirely understood but there is a monotonic change along the length of the LSO boule. The observed decay time of 34 ns in the sol-gel samples is well within this range. The production parameters can be better controlled in the sol-gel technique so it may be possible to produce LSO with more consistent timing parameters. The same is true for the scintillation light yield, which is also seen to vary, by as much as a factor of 2 in a single boule of Czochralski LSO.
Figure 6: Comparison of the Cs-137 energy spectra collected from a sol-gel crystal and a Czochralski crystal. System gain was the same for both measurements.
D. Light decay measurement One of the most valuable properties of LSO is its fast decay time relative to other high light yield inorganic scintillators. Decay times of the sol-gel grown LSO samples were compared to standard Czochralski grown crystals to ensure that this property is preserved. The samples were placed on a very fast PMT (RCA C3 1024) and irradiated with Na-22. There was no processing of the PMT signal, the signal was connected directly to a Tektronix TDS 380 digital oscilloscope and the pulse shapes were digitized and saved to a disk. In addition to the LSO samples, the decay time of a plastic scintillator (BC404, Bicron Inc., Newbury, OH) was measured to test the measuring system. BC404 has a quoted decay time of 1.8 ns. Figure 7 shows the digitized PMT pulses from the plastic scintillator, the standard LSO (Czochralski) and a sol-gel crystal sample. These curves were fit with single exponential
E. Gamma-ray spectral responses The experimental setup for the gamma-ray spectral response was identical to that used for the light output measurements. The spectral response of the LSO crystals to various gamma-ray sources (Cd-109: 22 keV and 88 keV; Am-241: 60 keV, Na-22: 51 1 keV, and Cs-137: 662 keV) was measured. The selected sources provide a range of photopeak energies from 22 to 662 keV. The linearity of the response, defined as the linearity of the photopeak position on the MCA versus gamma-ray energy, is illustrated in Figure 8 for a 1.5 x 2 x 0.85 mm3 sol-gel LSO crystal. These data show that the scintillation light intensity is quite linear over a wide range of energy.
1785
500 Y
=Na
v
400
;
f
5
(3
I
,
,
0
,,0' ,
a~ 300:
P
"'CS
Peak position
0
_ _ - _-Fit _
I
200 f
,
100
#
,
,+'
%J?
0
,
100 200 300 400 500 600 700
Peak position (channel number) Figure 8: Gamma response linearity of a sol-gel scintillating crystal.
Radiation sources Am-241, Cd-109, Na-22, and Cs-137 were used. Figure 9 shows the spectra used in the calibration of Figure 8. The % FWHM energy resolution is 64, 36, 19 and 15 at 22 keV, 60 keV, 5 11 keV and 662 keV, respectively. All spectra were acquired using a 1.5 ps shaping time on the spectroscopy amplifier.
In addition to good energy resolution, a high photofraction is very important in many applications. The photofraction is the ratio of photoelectric-to-photoelectric+scatter interactions measured by the detector system. The photofraction also depends on the dimensions of the crystal. In fact the light output of scintillators increases for shorter crystals due to the larger solid angle subtended to the PMT photocathode, but the photofraction decreases. The photofraction of the 1.5 x 2.0 x 0.85 mm3 Sol-Gel LSO was compared to a Czochralski crystal that was 2.0 x 1.7 mm2 in cross-section and measured at various thickness. The Sol-gel samples have very irregular shapes so the dimensions represent an average and the comparisons to the regular shape and polished Czochralski LSO are of a qualitative nature at this point. The best LdSi ratio achieved in the sol-gel process thus far is approximately 1:4. In Czochralski LSO (Lu2(Si04)O) there is a 2:l LdSi ratio whch greatly enhances the Z e ~ a n d hence the photofraction. The measured photofraction is plotted in Figure 10 as a function of scintillator thickness. The photofraction for a SolGel sample is about a factor of 5 below the trend suggested by the Czochralski results. This is a favorable comparison considering the difference in LdSi ratio and the smaller size of the Sol-Gel sample.
15000
12028820
20000
12000 .I
C
16000 f
I
no22 keV
* *
*
Na-2
g 12000 : 3 6 8000
*
%
0
a
i/\
511 keV
4000
%
3000 0
0
20
40
60
80
0
100
Energy (keV) 6000
Am-241
2500 f
-
2000
e
1500
. .:
0
1000
.
'E.
500
*
5000
60 keV
*
32
2
e
0
f e
3000 2000
1000
12M8821
1
A
1 I
0
CS-1371
0
I
:A 'j '
\
662keV
i
0 0 200 400 600 800 60 80 I00 Energy (keV) Energy (keV) Figure 9: Energy spectra for a sol-gel LSO scintillatingcrystal coupled to a PMT for Cd-109, Am-241, Na-22, and '3-137.
0
0
20
40
IOOC
1786
0.20
1
1
1
c 0
U
0
I
2 0.05 0.00
Czochralski
U Sol-Gel
5 0.0
1.o 2.0 Thickness (mm)
3.0
Figure 10. Photofraction = (events in photopeak) / (events in scatter spectrum) vs. Crystal Thickness for various thickness Czochralski LSO and one Sol-Gel sample.
IV. CONCLUSIONS A new approach has been developed using a lowtemperature crystal growth technology to produce scintillating LSO crystals. Light transparent polycrystalline LSO samples of a few mm3 in volume were grown and characterized by XRD, optical absorption, light decay measurement and gamma-ray spectral response. The performance of the sol-gel crystals was similar to that of the equivalent volume Czochralski grown crystals. The presented method of producing Lutetium silicate scintillating crystals offers many advantages over conventional techniques in terms of the better control of the process, significantly lower production costs and possibilities of growing crystals with custom sizes and shapes using casting techniques. Although, results are shown for Lutetium silicates the general method can be applied to other scintillating oxides. In order to increase LdSi ratio, reaction yields, size of the grown crystals, and improve gamma-ray response characteristics there is a further work required in the areas of optimization of precursor chemistry, gelation conditions, firing and casting procedures.
V. ACKNOWLEDGMENTS This work was supported by National Institutes of Health, Grant # 2R44GM57749.
VI. REFERENCES [l] R. Hofstadter, Phys. Rev., Vol. 74 (1948) 100 [2] C. L. Melcher and J. S. Schweitzer: A promising new scintillator: cerium-doped lutetium oxyorthsilicate. Nucl. Instr. Meth., A314 (1992) 212-214. [3] C.L. Melcher and J.S. Schweitzer, “Cerium-doped Lutetium Oxyorthosilicate: A Fast, Efficient New Scintillator”, IEEE Trans. Nucl. Sci., Vol. NS-39, No. 4. (1992) 502-505. [4] C.L. Melcher, R.A. Manente, C.A. Peterson, and J.S. Schweitzer, “Czochralski Growth of Rare Earth Oxyorthosilicate Single Crystals”, J. Crystal Growth, Vol. 128 (1993) 1001-1005. [5] C.L. Melcher, “Lutetium Orthosilicate Single Crystal Scintillator Detector”, US Patent No. 4,958,080 (1990). [6] R. Visser, C.L. Melcher, J.S. Schweitzer, H. Suzuki, and T.A. Tombrello, “Photostimulated Luminescence and Thermoluminescence of LSO Scintillators”, IEEE Trans. Nucl. Sci., Vol. NS-41, No. 4 (1994) 689-693. [7] J.D. Naud, T.A. Tombrello, C.L. Melcher, and J.S. Schweitzer, “The Role of Cerium Sites in the Scintillation Mechanism of LSO”, IEEE Trans. Nucl. Sci., Vol. NS-43, NO. 3 (1996) 1324-1328. [SI M.E. Casey, L. Eriksson, M. Schmand, M.S. Andreaco, M. Paulus, M. Dahlbom, and R. Nutt, “Investigation of LSO Crystals for High Spatial Resolution Positron Emission Tomography”, 1996 IEEE Nucl Sci. Symp. Conference Record, Vol. 2, 1029-1033. [9] T. Ludziejewski, K. Moszynska, M. Moszynski, D. Wolski, W. Klamra, N.O. Norlin, “Advantages and Limitation of LSO Scintillator in Nuclear Physics Experiments”, IEEE Trans. Nucl. Sci., Vol. NS-42, (1995), 328. [10]M. Moszynslu, T. Ludziejewski, D. Wolski, W. Klamra, M. Szawlowski, and M. Kapusta, “Subnanosecond Timing with Large Area Avalanche Photodiodes and LSO Scintillator”, IEEE Trans Nucl. Sci. Vol. NS-43, No. 3 (1996) 1298-1302. [ 111Brinker, C. Jeffrey., “Sol-Gel Science: The Physics and Chemistry of Sol-Gel,” C . Jeffrey Brinker, George W. Scherer. Boston: Academic Press, 1990. [ 121C. Cheng, Y, Xu, J.D. Mackenzie, Mat. Res. Soc. Symp. Proc. Vol271, pp.383-387 (1992) [ 131Melcher CL, Schmand M, Eriksson M, Eriksson L, Casey M, Nutt R, Lefaucher JL, Chai B, Scintillation Properties of LS0:Ce Boules, Presented at the 1998 IEEE Nucl. Sci. Symp., Toronto, CA. Nov 8-14, 1998.
